Assessment of international equity investment connectedness : portfolio diversification with respect to institutional and non-institutional investors

(1)

1

Assessment of International Equity Investment Connectedness:

Portfolio Diversification with Respect to Institutional and Non-institutional Investors

Agnė Galinskaitė

Supervisor at NHH: dr. Darya Yuferova Supervisor at VMU: dr. Asta Gaigalienė Master thesis in International Business and Finance

NORWEGIAN SCHOOL OF ECONOMICS VYTAUTAS MAGNUS UNIVERSITY

This thesis was written as a part of the Master of Science in Economics and Business Administration at NHH. Please note that neither the institution nor the examiners are responsible − through the approval of this thesis − for the theories and methods used, or results and conclusions drawn in this work.

Norwegian School of Economics Vytautas Magnus University Bergen, Kaunas, Spring, 2018

(2)

2

FOREWORD

I would like to thank the Norwegian School of Economics and the Vytautas Magnus University for giving me the possibility to obtain a double master degree with special thanks to Assoc. Prof. Dr.

Renata Legenzova. I also would like to thank my supervisors at Norwegian School of Economics Dr. Darya Yuferova and at Vytautas Magnus University Dr. Asta Gaigalienė for leading me through this time-consuming research. Finally, I thank all staff at Norwegian School of Economics and Vytautas Magnus University who helped me during the double master degree studies.

(3)

3

ABSTRACT

Author of diploma paper: Agnė Galinskaitė

Full title of diploma paper: Assessment of International Equity Investment Connectedness:

Porfolio Diversification with Respect to Institutional and Non-institutional Investors

Diploma paper advisors: Dr. Darya Yuferova, Norwegian School of Economics Dr. Asta Gaigalienė, Vytautas Magnus University Presented at: Norwegian School of Economics, Bergen, 2018

Vytautas Magnus University, Faculty of Economics and Management, Kaunas, 2018

Number of pages: 95

Number of tables: 25

Number of pictures: 10 Number of appendices: 44

International equity flows increased approximately five times from 2001 to 2016.

Therefore, stock market connectedness is increasing over time. It is necessary to assess international equity investment structure not only in general but also by disaggregating it by the type of investor as institutional and non-institutional investors have different characteristics and are important participants in financial markets. This thesis concentrates on international equity investment connectedness during growth and crisis periods with regard to institutional and non-institutional networks. Its structure is divided into three parts. The first part is dedicated to the literature review on differences between institutional and non-institutional investors, determinants of equity flows, methodologies used to assess stock market connectedness and contagion, its channels, structure of international equity investment network and its relevant measures. The second part covers the relevance, aim, logic of the research, steps, chosen evaluation methods, formulation of the research hypotheses and discussion of research limitations. The third part is devoted to the discussion of the results obtained analysing international equity investment connectedness with respect to institutional and non-institutional investors during growth and crisis periods.

It is found that institutional and non-institutional investors have different portfolio diversification practices. Institutional investors accounting for majority of equity flows form denser, more clustered, hierarchical and connected network. These differences persist even during crisis although both network are affected negatively. Even if there are significant differences between institutional and non-institutional networks during crisis, it does not induce relevant changes in the

(5)

5

structure of both networks. In addition, non-institutional investors are less vulnerable to financial crisis. However, both types of investors react negatively to increased stock market volatility during growth period. Besides stock market volatility, institutional investors, especially from central countries, diversify their portfolios in more countries when exchange rate volatility increases during growth period and contracts during crisis. Non-institutional investors, instead, do not consider exchange rate volatility as a significant risk factor. Finally, both types of investors invest more in countries with higher debt to GDP during growth period but withdraw their investments during crisis. This factor is the most relevant to non-institutional investors.

(6)

6

INTRODUCTION

The purpose of diversification is to get a higher profit with a lower risk. Hence, investors choose stocks in different sectors, currencies, countries that are in different stages of their economic life cycles. Due to globalisation processes and increased wealth of poor countries, total international equity investment flows increased from $5.2tn in 2001 to $24.6tn in 2016 (Coordinated Portfolio Investment Survey database [CPIS], 2018), which leads to more complex, less predictable financial networks with more sudden reactions to global events. According to Markowitz (1952), the best portfolios are constructed with low-correlated stocks. Since the global market risk is lower than that of separate countries, investments in different countries seem to be a good source of diversification.

However, it is discussed that higher financial integration is also related to enhanced risk of financial contagion. The financial crisis of 2007-2009 demonstrates that international equity investment flows are much more correlated than it was pronounced. Therefore, it raised uncertainty about the gains of portfolio diversification in foreign markets and inspire the discussion about the connectedness between equity markets. Now financial markets are much more connected than during crisis, when total equity investments accounted for $9.9tn. Therefore, the issue of stock market connectedness now is even more important than before.

In addition, investments in foreign equity markets are growing with regards to both institutional and non-institutional investors who are different in essence. It is commonly not agreed on type of investors stabilising financial markets (Barrot, Kaniel and Sraer, 2016; Zeng, 2016; Choi, Kedar-Levy and Yoo, 2015; Han, Zheng, Li and Yin, 2015; Foucault, Sraer and Thesmar, 2011;

Bohl, Brzeszczynski and Wilfling, 2009; Kaniel, Saar and Titman, 2008), however, institutional investors are the major market players (CPIS, 2017). Notwithstanding, non-institutional investors affect stock prices by increasing volatility in stock markets (Han et al., 2015; Foucault et al., 2011), changing direction of equity flows due to changes in risk aversion (Roque, Cortez, 2014) or destabilising equity markets by being too pessimistic or too optimistic about future prices of stocks (Fisher and Statman, 2000). Both institutional and non-institutional investors would choose assets not only with regards to their riskiness and returns but also for external factors such as country creditworthiness (Garg and Dua, 2014), market liquidity (Todea and Pleşoianu, 2013; Bekaert, Harvey, Lundblad and Siegel, 2011), political climate (Ahmed, 2017; Giofré, 2017; Erdogan, 2014) and others. Although average non-institutional investors underperform (Koestner, Loos, Meyer and Hackethal, 2017; Barrot et al., 2016; Calvet, Campbell and Sodini, 2009), nevertheless, they provide additional liquidity to financial market when institutional investors are restricted. Owing to differences between institutional and non-institutional investors, both international equity investment networks should be evaluated.

(7)

7

Taking into consideration that due to globalisation relations among international equity markets become more complex and less predictable, the network methodology is advantageous given that it assesses both direct and indirect financial links between equity markets within international equity investment network, its topology and detects links vulnerable to financial shocks in other markets. In regard to financial connectedness it is found that markets are linked to each other both directly and indirectly (Sh. Zhang, Wang, Liu and Wang, 2016; Chinazzi, Fagiolo, Reyes and Schiavo, 2013), there are central and peripheral financial markets (Chuluun, 2017;

Schiavo, Reyes, Fagiolo, 2010) and the probability of contagion depends on the network structure (Chuluun, 2017; Acemoglu, Ozdaglar and Tazbaz-Salehi, 2015; Elliott, Golub and Jackson, 2014;

Feroldi and Gaffeo, 2014; Chinazzi et al., 2013; Oatley, Winecoff Kindred, Pennock and Danzman, 2013). However, the studies do not analyse institutional and non-institutional investors separately.

Therefore, the problem of the thesis is how institutional and non-institutional investors portfolio diversification decisions affect formation of international equity investment connectedness?

The object of the master thesis is an international equity investment connectedness. The research is designed to achieve the aim which is to evaluate the influence of institutional and non- institutional investors portfolio diversification decisions on formation of international equity investment connectedness. In relation with the aim of the thesis, the main tasks are accomplished:

1. To analyse and synthesize financial literature about differences between institutional and non- institutional investors and determinants of equity flows.

2. To overview methodologies used to assess the connectedness and contagion among international equity markets, contagion channels and their results.

3. To develop the methodological background for assessment of international equity investment connectedness with respect to institutional and non-institutional investors.

4. To analyse general differences in international equity investment connectedness with respect to institutional and non-institutional investors, differences within and between international equity investment networks during growth and crisis periods, stock market and country riskiness impact on international equity investment networks with regard to institutional and non- institutional investors during growth and crisis periods.

5. To discuss the results with findings of other researchers, implications and suggest possible research extensions.

Methods and sources: the qualitative analysis is conducted analysing and summarizing findings in the scientific literature. Sources for qualitative analysis are gathered from Jstor, Science Direct, EBSCO Business Source Complete, Emerald Insight, Springer Link, Cambridge Journals Online, SSRN and International Monetary Fund (IMF). Assessing the influence of institutional and non-institutional investors portfolio diversification decisions on formation of international equity

(8)

8

investment connectedness, the quantitative analysis is based on formation of network matrices, calculation of aggregate and node specific indices, running OLS and random/fixed effects regressions and comparing the results for different networks. The data for quantitative analysis is gathered from CPIS provided by IMF, World Bank and Thomson Reuters Eikon database. The research is conducted using Microsoft Office 2016, packages for network analysis Netminer and Gephi and statistical analysis software STATA.

The structure of this thesis is divided into three parts. In the first part I conduct literature review which enables to formulate research hypothesis. In literature review I discuss about the differences between institutional and non-institutional investors from the point of view of participation in stock markets and investor characteristics. Then, I overview which factors, according previous literature, are significant investing abroad. Later, I analyse and compare methodologies used to assess stock market connectedness, contagion, its channels and their results.

Finally, I present findings of previous literature on financial connectedness and contagion using network methodology, summarize the usage of network structure measures by other researchers. In the second part I explain the relevance, aim, logic of the research, steps, chosen evaluation methods, formulate the research hypotheses and discuss research limitations. The third part is devoted to the discussion of the results obtained analysing the influence of institutional and non- institutional investors portfolio diversification decisions on formation of international equity investment connectedness. Firstly, I present general tendencies of international equity investment connectedness with respect to institutional and non-institutional investors and compare them.

Secondly, I analyse how the gap in international equity investment connectedness with regard to institutional and non-institutional investors differ during crisis and structural changes caused by crisis within international equity investment networks with respect to institutional and non- institutional investors (later institutional and non-institutional networks). Thirdly, I evaluate stock market and country risk impact on the structure of both networks during growth and crisis periods.

Finally, I compare the results with findings of other researchers and discuss their possible implications.

Findings of the conducted research reveal that institutional and non-institutional investors form different international equity investment networks: institutional investors have larger, denser, more clustered and hierarchical network. Clusters are not only denser but also consist of different countries. In addition, institutional investors invest in more distant countries, for example, Africa, therefore, the results are in line with findings of Roque and Cortez (2014) that non-institutional investors are more linked to neighbour financial markets. In addition, I find that the United States, the United Kingdom, Italy and France are important intermediaries connecting separate clusters in both networks. Even though non-institutional investors obtain less partnerships and generate less

(9)

9

equity flows, they are also vulnerable to financial crisis given that separate clusters and communities in the network are connected through direct and indirect links from central countries.

Hence, both networks have characteristics of hierarchical and flat networks. Therefore, the results of Feroldi and Gaffeo (2014) analysing total financial portfolio and Schiavo et al. (2010) analysing international equity investments that both types of investment networks are hierarchical and consist of G7 countries are close to my findings, nevertheless, central countries in institutional and non- institutional networks partially differ.

Financial crisis makes twofold impact on the gap between connectedness measures in institutional and non-institutional networks. The differences in relations, determined by the number of connections, increase, while differences in relations, determined by equity flows, decrease suggesting that flows in institutional network are distributed in a larger array of countries and vice versa in non-institutional network. However, crisis does not induce significant internal changes in any of networks. The changes occurring in institutional network are due to ordinary countries which get and invest less funds because institutional investors from central countries do not change their investment strategies. These findings are in line with results of Chinazzi et al. (2013) that central countries are less affected by crisis. Non-institutional investors, instead, maintain the volume of investments but invest in fewer countries. Therefore, these findings complement the findings of Hoffmann et al. (2013) and Roque and Cortez (2014) that non-institutional investors prefer more transparent financial markets during crisis.

I also find that both institutional and non-institutional investors from ordinary and central countries react negatively to market riskiness. In addition, institutional investors from central countries diversify their portfolios in more countries during growth period and in less countries during crisis when exchange rate volatility increases. Non-institutional investors are not affected by exchange rate volatility in general. Hence, the statement of Ang and Bekaert (2002) that international diversification has a positive value for international portfolios, if the currency exchange rate risk is hedged, is relevant only to institutional investors. Moreover, an increase in public debt to GDP during growth period is a positive sign for both investors, but during crisis countries which have less sound economics lose significant volume of investments from institutional investors and partnerships with non-institutional investors. Findings of Bekaert, Ehrmann, Fratzscher and Mehl (2014) that during crisis investors care more about macroeconomic factors than stock market riskiness are possible to confirm but during growth period investors, especially non-institutional, take into account the level of public debt to GDP. Finally, the research results reveal that institutional investors considering public debt to GDP induce substantial losses in ordinary countries and gains in central countries summing equity inflows during growth and crisis periods.

(10)

10

I. THEORETICAL ASPECTS OF FINANCIAL CONNECTEDNESS IN INTERNATIONAL EQUITY INVESTMENT NETWORK

This chapter is dedicated to the review and discussion in regard of financial literature. The relevance and differences between institutional and non-institutional investors are discussed in section 1.1. Section 1.2 summarizes factors determining international equity investment flows.

Moreover, section 1.3 provides the conceptual clarity of financial connectedness, contagion and its channels (for vocabulary of terms see Appendix 1). Finally, network structure, its impact to financial connectedness and contagion, and its mostly used measures are summarized in section 1.4.

1.1 Differences between institutional and non-institutional investors

Although the scope of portfolio diversification is common for all types of investors, their behaviour and target countries could vary under the different macroeconomic conditions. Basically, investors can be distinguished into four groups: institutional, non-institutional, monetary authorities and non-profit organisations controlled by government. However, the contribution of monetary authorities and government-related units to total equity investment flows is less than 1% (Appendix 2). Hence, many authors analyse investments only with regard to institutional and non-institutional investors (Li, Rhee and Wang, 2017; Choi et al., 2015; Roque, Cortez, 2014).

There is no single definition of terms “institutional investor” and “non-institutional investor” because it varies based on the emphasized characteristic. However, an institutional investor is a legal entity (Çelik and Isaksson, 2014) and often is treated as an informed investor, while a non-institutional investor is regarded as a biased investor (Kaniel et al., 2008). They have different roles in the financial markets and possess specific characteristics (Table 1). Institutional investors are entities which invest funds on behalf of their investors and are regulated by monetary authorities. Non-institutional investors, who are physical people or other than professional investment companies, do not rise as many funds as institutional investors (Ivković, Sialm and Weisbenner, 2008), do not have specific knowledge about capital markets and do not have or not enough experience in investment (Stiglitz, 2003). The same results are found analysing purchase of stocks before and after bad mergers (Han and Chung, 2013), forecasts of stock returns (Choi and Sias, 2012) and ownership influence on bid-ask spreads (Schnatterly, Shaw and Jennings, 2008).

Although non-institutional investors are relatively unskilled, their performance varies highly within the group. Bailey, Kumar and Ng (2008) come to the conclusion that wealthier and more experienced individual investors are more successful in foreign markets than others. Calvet, Campbell and Sodini (2009) suggest that the reason is investments in riskier assets. Koestner, Loos, Meyer and Hackethal (2017) agree with a positive correlation between experience and payoff –

(11)

11

having more practice, non-institutional investors make better investment decisions. Therefore, non- institutional investors compared to institutional investors are rather unskilled and lack of funds but the differentials decrease when non-institutional investors become more experienced.

Table 1

Differences between institutional and non-institutional investors

Feature Sub-feature Institutional Non-institutional

Quality of subject

Availabililty of funds Higher Lower

Financial literacy Skilled Relatively unskilled

Legal restrictions

Use of intermediaries Do not use Use

Access to primary and

secondary markets Primary and secondary Secondary

Taxes Lower Higher

Engagement in stock market

Stock preferences Dividend / growth Growth / dividend

Market stabilisation Noise traders / contrarians Contrarians / noise traders

Share in a stock market ~82.5% ~17.5%

Investment bias

Risk aversion Relatively less risk-averse Relatively more risk-averse

Overconfidence Both can be overconfident

Equity home bias Non-biased Biased

Herding Relatively less Herd

Investment abroad

Market development - Prefer developed markets

Market transparency Less transparent More transparent

Risk diversification Higher Lower

Geographical distance More distant markets Closer markets

Common stock exchange - Prefer

Portfolio diversification More diversified Less diversified Note: done by author based CPIS database (2017); Choi et al., (2015); Roque and Cortez (2014); Giofré (2013); Lai et al. (2013); Chiang et al. (2012); Jain (2007); Campbell (2006); Allen, Bernardo and Welch (2000).

Given that non-institutional investors are not professional investors, their participation in financial markets is restricted by law. Institutional investors are referred as financial specialists because they make investments using funds of other subjects. Non-institutional investors, instead, often consult with financial advisors before making investment decisions (Campbell, 2006), therefore, advised non-institutional investors perform better (Gaudecker, 2015). Kramer (2012) and Bhttacharya, Hackethal, Kaesler, Loos and Meyer (2012) arrive at different results – Bhttacharya et al. (2012) conclude that differences between investment performance of advised non-institutional investors and self-leading non-institutional investors are not significant. Kramer (2012) findings reveal significant differences between advised and non-advised non-institutional investors by explaining this phenomenon as a conflict of interests between non-institutional investors and financial advisors. However, portfolios of advised non-institutional investors are diversified better.

Another legal restriction occurs when institutional investors access primary and secondary markets because certain types of institutional investors, for example, investment banks, serve as underwriters in the primary market (Bonaventura, Giudici and Vismara, 2017), while non- institutional investors can participate only in the secondary markets. Non-institutional investors are determined not only by lower accessibility to financial markets but also by dividend tax

(12)

12

disadvantage (Kawano, 2014; Allen, Bernardo and Welch, 2000). Owing to tax advantage for institutional investors (Kawano, 2014), they prefer large-cap dividend-paying stocks, while non- institutional investors prefer small-cap growth stocks with high leverage (Bae, Min and Jung, 2011;

Allen et al., 2000). Kawano (2014) concludes that the choices of non-institutional investors are influenced by dividend taxes – with a dividend tax reduction by 1% long-term profitability to non- institutional investors increases by 0.04%. Therefore, non-institutional investors choose more dividend stocks. However, non-institutional investors perform worse than institutional investors – on average, they underperform the market even before taxes (Barber and Odean, 2013). This result might be caused by many factors: illiteracy, scarce funds, transaction costs, investment biases and others.

Black (1986) suggests that noise trading is related to investment underperformance due to false thinking that noise contains information. Although average non-institutional investor incurs losses, in general, they provide liquidity to financial markets. According Foucault et al. (2011), along with liquidity provided by non-institutional investors, comes higher volatility destabilising the financial markets. Thus, non-institutional investors are important actors, although their equity market share accounts for approximately 17.5% (CPIS database, 2017). Recently, scientists highlight that both institutional and non-institutional investors could have a role of market stabilisation. Barrot et al. (2016) find that non-institutional investors offer additional liquidity during financial crisis when institutional investors are restricted. Institutional and non-institutional investors are different market agents because non-institutional investors underperform market (Barrot et al., 2016; Griffin, Harris and Topaloglu, 2003) selling assets before increase in their value, while institutional investors buy them (Griffin et al., 2003). Zeng (2016) analysing trading of institutional investors in the United States concludes that the relationship between overvalued stocks and institutional holdings is statistically significant. Analysing S&P 500 and WIG 20 Bohl et al. (2009) contribute with different findings – institutional investors have equity market stabilising roles in the United States and Poland. Han et al. (2015) analyse the role of foreign and local institutional investors in China and find that foreign institutional investors stabilise equity market, while local institutional investors increase stock market volatility. Therefore, origin of investor also matters.

Foreign investors form international equity investment networks. Their investment decisions are based not only on knowledge but also on preferences. Risk-aversion determines willingness to risk. Therefore, it strongly affects investment network because with higher risk- aversion certain countries and type of assets are avoided. For example, low return stocks are not attractive when the sentiments of investors are high (Baker and Wurgler, 2006). Luchtenberg and Seiler (2014) find that risk-aversion does not depend on the type of investor – it is, rather, a human

(13)

13

phenomenon: if the value of the stock increases, both institutional and non-institutional investors are risk averse; when the value of the stock decreases, both types of investors prefer to risk.

However, non-institutional investors are more risk-averse (Basak and Pavlova, 2013), especially households which are risk averse in such a manner that portfolios become underdiversified (Gaudecker, 2015; Campbell, 2006; Barber and Odean, 2000).

Non-institutional investors perform worse owing to other behavioural biases. For example, according Barber and Odean (2013) and Ivković et al. (2008), the under-diversification can be caused by asymmetric information, overconfidence and familiarity bias. Other reason, why non- institutional investors invest in few stocks, in the sight of Ivković et al. (2008), is a small amount of money in hands. Overconfidence can occur for different reasons, for example, in belief that investor can invest better than average (Barber and Odean, 2013). Non-institutional investors are more overconfident than institutional investors (Liu, Chuang, Huang and Chen, 2016) due to lack of knowledge about financial instruments (Barber and Odean, 2013). In addition, the overconfidence varies depending on the market liquidity and volatility: in more liquid (more central in equity investment network) and less volatile markets both institutional and non-institutional investors are less overconfident.

Overconfidence is also associated with home bias. In Døskeland and Hvide (2011) terms, non-institutional investors fail not only to invest in close to their profession stocks but also in other local stocks. Seasholes and Zhu (2010) find that non-institutional investors perform worse when invest in local firms. However, not only non-institutional investors are home-biased. In the countries where uncertainty avoidance is high, institutional investors are home-biased (Choi, Fedenia, Skiba and Sokolyk, 2017). In addition, institutional investors from countries which are less tolerant to uncertainty and have higher cultural distance under-diversify their investment portfolios abroad (Anderson, Fedenia, Hirschey and Skiba, 2011).

Herding is related to the equity network formation and its changes in crisis periods. It is more pronounced among non-institutional investors, especially when an investor had a negative experience without herding (Merli and Roger, 2013). However, Basak and Makarov (2014) and Sias (2004) agree with the fact that even institutional investors before investing evaluate strategies of their competitors but arrive at different conclusions: according Basak and Makarov (2014), institutional investors choose strategies with different directions; according Sias (2004), institutional investors seek investments of other investors. Barber, Odean and Zhu (2009) come up with the same conclusions regarding non-institutional investors. However, Li et al. (2017) find that non-institutional investors are affected more by public information. The dispersion of investments is lower for non-institutional investors and is related to market movement. In addition, non-

(14)

14

institutional investors herd buying in growth periods, while selling only during crisis. Therefore, crisis is important forming institutional and non-institutional networks.

Non-institutional investors have less diversified portfolios than institutional due to their features such as biasness, lack of knowledge and funds. Institutional investors are less sensitive to transparency in foreign equity markets (Roque, Cortez, 2014; Giofré, 2013), existence of common stock exchange (Giofré, 2013), physical distance (Roque, Cortez, 2014) and market development (Roque, Cortez, 2014). As a result, institutional investors have more diversified investment portfolios and risk. In addition, institutional and non-institutional investors react differently to financial crisis: non-institutional investors choose equity investment in more developed countries which are more transparent, while institutional investors prefer investing in stocks allowing better portfolio diversification (Roque, Cortez, 2014). Hoffmann, Post and Pennings (2013) and Roque and Cortez (2014) find that non-institutional investors do not change their equity investment strategies with exception of common currency and investor protection which become irrelevant explanatory variables during crisis.

Based on their characteristics, investors fall into certain groups and, consequently, incur limitations forming personal preferences for portfolio diversification in foreign equity markets.

Although institutional and non-institutional investors differ in their qualities, participation in equity markets and preferences for specific stocks and countries, an analysis of the paper is restricted to the differences arising from investments in foreign markets, summarized in the last section of Table 1. However, given that the global equity market is wide and various, there are many global, bilateral or country-specific factors affecting investment choices in other equity markets.

1.2 Factors determining international equity flows

When investors construct their portfolios, the primary concern is to diversify risk obtaining the highest possible returns. Nevertheless, the direction of investments is induced by many factors which can be grouped according to their origin. The thesis classifies foreign investment determinants to market risk, barriers and development (Table 2), information costs, familiarity and bilateral links (Table 3) and quality of legal and financial system (Table 4) factors.

Beginning with the market riskiness, the optimal portfolio theory is introduced by Markowitz (1952) demonstrating that investment portfolio should be diversified choosing stocks from different industries, e.g. stocks which have lower return correlation/covariance (Table 2). In addition, besides equity returns, investors take account of dividend yields – non-institutional investors prefer growth stocks due to tax disadvantages (Kawano, 2014; Bae, Min and Jung, 2011).

Given that the stock markets are determined by different level of riskiness, investors consider equity market and country specific risk factors. Such risks include stock market (Cai, Mobarek and Zhang,

(15)

15

2017) and exchange rate (Giofré, 2017) volatilities, market liquidity (Bekaert et al., 2011) and sovereign risk (Bekaert et al., 2014). Foucault et al. (2011) and Han et al. (2015) reveal that increase in stock market volatility is influenced by both institutional and non-institutional investors.

Therefore, the relationship between stock market volatility and investing preferences should be bidirectional.

Table 2

Determinants of international equity investment flows: diversification, risk, returns and barriers

Class Determinant Source (year)

Risk

diversification Return Correlation Karolyi et al. (2015); Roque and Cortez (2014); Garg and Dua (2014); Bekaert at al. (2009); Markowitz (1952)

Foreign market risk

Global risk Cai et al. (2017); Bekaert et al. (2011) Stock market volatility Cai et al. (2017)

Exchange rate (volatility)

Cai et al. (2017); Giofré (2017); Kanas and Karkalakos (2017);

Abid et al. (2014); Garg and Dua (2014); Gyntelberg et al. (2014);

Bekaert (1995) Country risk Bekaert et al. (2014)

Market liquidity Todea and Pleşoianu (2013); Bekaert et al. (2011); Bekaert (1995) Returns Equity returns Stepanyan (2017); Al-Khouri (2015); Karolyi et al. (2015); Garg

and Dua (2014); Roque and Cortez (2014) Dividend yield Cai et al. (2017)

Foreign market size /

development

GDP* Muzur et al. (2015); Roque and Cortez (2014); Qian and Steiner (2014)

GDP per capita Giofré (2017); Karolyi et al. (2015); Erdogan (2014);

GDP growth rate Mobarek et al. (2016); Abid et al. (2014); Aggarwal et al. (2012);

Bekaert (1995) Interest (real) rate

differential,

Inflation (volatility)

Stepanyan (2017); Cai et al. (2017); Mobarek et al. (2016); Mollah et al. (2016); Luchtenberg and Vu (2015); Abid et al. (2014);

Erdogan (2014); Garg and Dua (2014); Bekaert (1995) Market capitalisation

Baumöhl et al. (2018); Mobarek et al. (2016); Luchtenberg and Vu (2015); Erdogan (2014); Roque and Cortez (2014); Qian and Steiner (2014); Kuvvet (2013); Todea and Pleşoianu (2013); Bekaert (1995) Market turnover Cai et al. (2017); Karolyi et al. (2015)

Unemployment rate Bekaert et al. (2014)

Investment barriers

Capital controls Giofré (2017); Karolyi et al. (2015); Erdogan (2014); Qian and Steiner (2014); Giofré (2014)

Transaction costs /

taxes Karolyi et al. (2015); Todea and Pleşoianu (2013); Bekaert (1995) Equity market

openness Abid et al. (2014); Qian and Steiner (2014); Bekaert et al. (2011) Note: * Bekaert et al. (2014) also find statistically significant government budget and current account factors.

Luchtenberg and Vu (2015) – industrial production; Cai et al. (2017) – currency reserves.

Turning to exchange rate volatility, it is found to have a negative impact on investment diversification (Baumöhl et al., 2018), nevertheless, investments can be successful if it is hedged (Ang and Bekaert, 2002). Hedging of exchange rate volatility is irrelevant when the assets are from illiquid stock markets inasmuch as they enhance the riskiness of an asset due to low trading frequency. As a result, investors prefer more liquid equity markets (Bekaert et al., 2011; Bekaert, 1995). In addition, country risk is also a relevant factor because it comprises sovereign, political and other risks that enhance systemic risk. Bekaert et al. (2014) point out that sovereign risk should be

(16)

16

considered especially during crisis owing to changes in risk aversion of investors based on macroeconomic fundamentals. Moreover, Qian and Steiner (2014) find that higher central bank reserves have a positive influence on foreign equity investments because higher reserves reduce exchange rate risk and risk premium required by investors.

Nonetheless, financial soundness is not the only factor pertinent to foreign investors – the size (determined by GDP) of the country is also significant. Size, according to Muzur et al. (2015), is relevant as the bigger countries have more possibilities to invest abroad and attract foreign investments. In addition, Bekaert and Hoerova (2016) highlight that country size is correlated (positively in the United States and negatively in Germany) with a risk aversion. Therefore, country of equity holders also should be considered. GDP per capita, instead, is connected with a country’s strength/development which is positively related to foreign investments (Giofré, 2017). Country development can be also expressed by its equity market capitalisation – when the capitalisation is low, illiquid market demotivates foreign investors to diversify their portfolios in the country (Erdogan, 2014; Bekaert et al., 2011). While low market capitalisation pushes foreign investors from equity markets indirectly, government policy through investment barriers such as high capital controls (Giofré, 2017), taxes (Karolyi et al., 2015) and low equity market openness (Bekaert et al., 2011) directly and negatively influence foreign equity inflows. However, due to globalisation, investment barriers diminish, for instance, Bekaert et al. (2011) reveal that such highly controllable sector as banking now is the most integrated in the world market.

Although the financial markets are becoming more integrated, the informational disparity persists (Table 3). Hence, there is a broad literature on factors which can be also called bilateral:

common language (Giofré, 2017), religion (Hellmanzik and Schmitz, 2017), legal system (Giofré, 2017), common currency (Roque, Cortez, 2014) that lessens information costs, and cultural distance (Roque, Cortez, 2014) factors which reduce familiarity with other markets and increase information costs. More integrated markets or markets that were integrated in the past are more likely to have colonial links (Karolyi, Ng and Prasad, 2015), bilateral trade (Erdogan, 2014), FDI (Baumöhl, Kočenda, Lyócsa and Vżrost, 2018), migration (Hellmanzik and Schmitz, 2017) and, as a result, bilateral equity flows. Hence, countries that develop bidirectional trade are also engaged in bilateral foreign equity investments (Muzur, Suesse and Krivitsky, 2015; Qian, Steiner, 2014). Due to globalisation and increased financial market openness, equity flows are increasing in both developed and developing markets, still, giving the priority to the countries which had relations in the past regarding FDI flows, subsidiaries and partnerships. Karolyi et al. (2015) explain it by obtained information cost advantages that are higher for developed countries. In addition, foreign investments are determined not only by cultural proximity and familiarity but also by geographical distance which induces familiarly bias. According to Roque and Cortez (2014), non-institutional

(17)

17

investors prefer investments in culturally closer and less geographically distant countries. Even though geographical distance is reduced by technological advantages (Hellmanzik, Schmitz, 2016), the time zone differences in trading keep existing (Erdogan, 2014).

Table 3

Determinants of international equity investment flows: information costs and familiarity

Determinant Source (year)

Common language Giofré (2017); Hellmanzik and Schmitz (2017); Karolyi et al. (2015); Erdogan (2014); Giofré (2014); Roque and Cortez (2014); Aggarwal et al. (2012)

Common religion Hellmanzik and Schmitz (2017); Mobarek et al. (2016); Roque and Cortez (2014);

Aggarwal et al. (2012) Common legal system

origin

Giofré (2017); Hellmanzik and Schmitz (2017); Erdogan (2014); Giofré (2014);

Aggarwal et al. (2012)

Cultural distance Mobarek et al. (2016); Roque and Cortez (2014); Aggarwal et al. (2012) Currency union Giofré (2017); Erdogan (2014); Giofré (2014); Roque and Cortez (2014);

Colonial links Giofré (2017); Hellmanzik and Schmitz (2017); Karolyi et al. (2015); Erdogan (2014); Giofré (2014)

Bilateral trade/net trade/exports/

imports/trade openness

Baumöhl et al. (2018); Cai et al. (2017); Mobarek et al. (2016); Muzur et al. (2015);

Luchtenberg and Vu (2015); Karolyi et al. (2015); Erdogan (2014); Roque and Cortez (2014); Kuvvet (2013)

FDI Baumöhl et al. (2018); Karolyi et al. (2015); Qian and Steiner (2014) Bilateral migration Hellmanzik and Schmitz (2017)

Geographical/virtual distance

Giofré (2017); Hellmanzik and Schmitz (2017); Karolyi et al. (2015); Erdogan (2014); Giofré (2014); Roque and Cortez (2014); Aggarwal et al. (2012) Common border Giofré (2017); Karolyi et al. (2015); Erdogan (2014); Giofré (2014) Time zone Hellmanzik and Schmitz (2017); Erdogan (2014)

Note: done by author.

The uncertainty of legal and political system highly affects risk-averse investors. For example, Roque and Cortez (2014) conclude that non-institutional investors prefer investing in more transparent countries. Transparency itself is related to stability and legal protection in the country (see Table 4). It is found that the standard of living and the level of corruption are inversely correlated (Lučić, Radišić and Dobromirov, 2016) and the efficiency of judicial, legal system, political stability, investor protection is higher and the expropriation risk is lower in more developed countries (Giofré, 2017, 2014). Wu, Li and Selover (2012) analyse how free flow of information and public trust affect international financial flows and arrive at the conclusion that countries with higher information availability to public also have higher public trust and investor protection. Accounting standards, which are related to the rule of law, should be taken into consideration due to difficulties that arise comparing performance of the companies which annual reports are based on different accounting standards (KPMG, 2015). Although accounting standards are not related to the level of development, other factors such as higher transparency, stability and investment protection determine advance of equity markets distinguishing high and low investment risk countries.

(18)

18 Table 4

Determinants of international equity investment flows: transparency, stability and legal protection

Determinant Source (year)

Corruption Giofré (2017); Jain et al. (2017); Mollah et al. (2016); Giofré (2014); Roque and Cortez (2014); Qian and Steiner (2014); Bekaert et al. (2011)

Judicial system efficiency Giofré (2017); Jain et al. (2017); Giofré (2014);

Legal system Giofré (2017); Erdogan (2014); Qian and Steiner (2014); Giofré (2014); Kuvvet (2013); Bekaert et al. (2011)

Investor protection Bao and Lewellyn (2017); Stepanyan (2017); Roque and Cortez (2014); Kuvvet (2013); Aggarwal et al. (2012); Giannetti and Koskinen (2009); Bekaert (1995) Expropriation risk Giofré (2017); Stepanyan (2017); Giofré (2014); Kuvvet (2013)

Accounting standards Giofré (2017); Giofré (2014); Bekaert (1995) Information availability Bekaert (1995)

Political stability Giofré (2017); Bekaert et al. (2014); Erdogan (2014); Giofré (2014); Bekaert (1995)

Government effectiveness Karolyi et al. (2015)

Rule of law/legal origin Giofré (2017); Karolyi et al. (2015);Giofré (2014); Bekaert et al. (2011) Note: done by author.

Analysing international equity investment flows, it is important to evaluate not only common foreign investment determinants such as currency, country, exchange rate risks, size of the country, diversification and return but also information costs and costs that arise due to the different cultures, languages, legal systems, physical distance and exposure to corruption, expropriation and political instability. Nevertheless, the research concentrates on the most commonly analysed stock market and country risk factors such as stock market volatility, exchange rate volatility and public debt to GDP. Institutional and non-institutional investors considering many factors, including risk factors, diversify their portfolios in foreign equity markets. As a consequence, they unintentionally form certain financial structures determined by certain connectedness characteristics which could become a source of direct or/and indirect financial contagion.

1.3 Concept and methodologies assessing international equity investment connectedness and contagion

Given that the financial contagion is a result of increased financial connectedness between equity markets, the methodologies used to assess both financial connectedness and contagion are similar. Nevertheless, there is a number of different methodologies unveiling alternate aspects of unilateral, bilateral and multilateral financial relationships, therefore, in the first part of this section the concept and methodologies used to assess the financial connectedness and contagion and results are discussed. As financial shocks can spread in different ways, the second part covers a review of financial literature which discusses about possible contagion channels.

1.3.1 Methodologies used to assess financial connectedness and contagion

The concept of financial connectedness in equity markets is closely related to stock market liberalization and globalisation. Globalisation effect on financial markets and their efficiency are

(19)

19

twofold. One stream of literature argues that countries, which open equity markets to foreign investors, benefit from financial liberalisation because it induces foreign investment flows (Fuchs- Schündeln and Funke, 2003), market efficiency (Bekaert et al., 2011), economic and productivity growth (Gehringer, 2012). When equity markets are open to foreign investments, investors can exploit the possibility to diversify their portfolios in different financial markets with reduced cost of equity. Other stream of literature see globalisation as a possible source of non-diversifiable risk making equity markets prone to financial shocks in other countries. However, scientists argue whether global financial crisis of 2007-2009 was induced by increased global risk presenting different results. Indeed, alternative methodologies, often, generate controversial results.

Nevertheless, different results do not indicate that one or other technique is unreliable – they are used to unveil different aspects of a problem. Methodologies used to assess financial connectedness and contagion are similar (Table 5).

Table 5

Methods used to analyse financial connectedness and integration

Method Articles about connectedness Articles about contagion

CAPM

CAPM: Chaudhary (2016); ICAPM: Abid et al. (2014);

Guesmi and Nguyen (2014); Guesmi et al. (2014);

Guesmi and Teulon (2014); Teulon et al. (2014); Berger and Pozzi (2013); Guesmi et al. (2013); Guesmi and Nguyen (2011); Bekaert (1995); Bekaert and Harvey (1995); IAPM: Carrieri et al. (2013)

ICAPM: Guesmi et al. (2013)

Factor Nardo et al. (2017)

Bae and Zhang (2015); Bekaert et al. (2014); Baele and Inghelbrecht (2010); Bekaert et al. (2005) Correlation,

Wavelet

Correlation: Nardo et al. (2017); Lucey and Zhang (2010); Baele (2005); Solnik et al. (1996); Wavelet:

Shah and Deo (2016); Graham et al. (2013); Rua and Nunes (2009)

Solnik et al. (1996)

Cointegration, VAR

Cointegration: Caporale et al. (2016); Lagoarde-Segot and Lucey (2007); Palac-McMiken (1997); VAR: Baele and Soriano (2010); Bekaert et al. (2002)

VAR: Cai et al. (2017); Forbes and Rigobon (2002); Royen (2002); GVAR: Beirne and Gieck (2014)

GARCH, DCC

GARCH: Berger and Pozzi (2013); Baele (2005);

ARCH-M: Carrieri et al. (2007); VAR-GARCH: Dutta (2018); DCC-GARCH: Guesmi and Nguyen (2014);

Guesmi and Teulon (2014); Guesmi, Moisseron and Teulon (2014); Guesmi et al. (2013); Guesmi and Nguyen (2011); GDC-GARCH: Abid et al. (2014);

ARFIMA-GARCH: Lyocsa, Vyrost and Baumöhl (2017); c-DCC-FIAPARCH: Teulon et al. (2014);

DCC: You and Daigler (2010); VAR-DCC: Al Rahahleh et al. (2017); VECM-DCC: Al Rahahleh et al. (2017)

GARCH: Baumöhl et al. (2018)*;

Billio and Caporin (2010); Baele and Inghelbrecht (2009); DCC- GARCH: Guesmi et al. (2013);

ADCC-GARCH: Mensi et al.

(2017); EGARCH: Khallouli and Sandretto (2012); DCC-MIDAS:

Mobarek et al. (2016)

Network

Chuluun (2017); Diebold and Yilmaz (2015); Sh. Zhang et al. (2016); Naitram (2014); Chinazzi et al. (2013);

Diebold and Yilmaz (2013)

Diebold and Yilmaz (2015);

Minoiu et al. (2015); Chinazzi et al. (2013)

Note: Baumöhl et al. (2018) use GARCH model and 9 its derivations.

Capital asset pricing model (CAPM) is a single factor model, which decomposes return on equity separating risk free rate and risk premium. This model assesses the asset riskiness evaluating

(20)

20

what risk premium should be given to investor investing in a certain market when government 10- year bonds are assumed to have risk free rate. CAPM can be calculated for segmented (separate markets), integrated (international CAPM which includes currency exchange risk) or partially integrated markets. However, integration of equity markets is a time-consuming process, hence, Bekaert and Harvey (1995) augment static CAPM model by allowing degree of integration to change over time. They find twofold results: such countries as Colombia, Jordan, Korea and Malaysia are integrated, while Chile, Greece, India, Mexico, Nigeria, Taiwan, Thailand and Zimbabwe are not. Nevertheless, these results should be interpreted with caution because Chile, Greece, Mexico, Korea and Zimbabwe failed in specification tests. Berger and Pozzi (2013) analyse equity time-varying market integration of the United States, Japan, Germany, France and the United Kingdom for period 1970-2011 and conclude that financial integration increased among all countries except Japan. Notwithstanding increasing integration in a long-term period, integration is not unidirectional – there are periods when country-specific shocks lead to disintegration. Guesmi, Moisseron and Teulon (2014) using ICAPM, find that MENA countries are regionally integrated.

Demir and Coşkun Kaderli (2015) analyse which models are more suitable to assess the cost of equity in Turkey and arrive at the conclusion that local CAPM and other local measures that are not suitable, hence, models which evaluate market integration, such as world CAPM, are better indicators. The same results are found by Chaudhary (2016)¹, Abid, Kaabia and Guesmi (2014)², Guesmi and Nguyen (2014)³, Guesmi and Teulon (2014)⁴. Analysing integration of not only regional but also of global equity markets, Guesmi and Nguyen (2011)⁵ find that countries are integrated regionally but are quite segmented globally, especially emerging markets. As Bekaert and Harvey (1995) notice, model can provide biased results. Given that the CAPM is a one-factor model, lack of other important explanators can influence the reliability of results. This ICAPM drawback is offset by factor models. Bekaert, Harvey and Ng use two-factor (2005), while Bekaert et al. (2014) multi-factor models to estimate equity market contagion. This model is useful assessing determinants of equity flows. Because the model is time-variant, it can estimate bilateral connectedness and contagion. However, neither time-variant CAPM nor factor models do not consider multilateral relationships in complex financial network.

The simplest co-movement method is correlation between stock prices/returns. Drawbacks of the unconditional correlation, which is measured calculating Pearson correlation, are that it treats historical data equally, in consequence, the results are highly affected by outliers, thus, correlation between equity markets could be higher than actually it is (Nardo, Ndacyayisenga, Papanagiotou,

1 Chaudary (2016) analyse India and the United States.

2 Abid, Kaabia and Guesmi (2014) analyse Indonesia, Malaysia, Singapore, Sri Lanka and Thailand.

3 Guesmi and Nguyen (2014) analyse Czech Republic, Greece, Poland and Romania.

4 Guesmi and Teulon (2014) analyse Egypt, Israel, Jordan and Turkey.

5 Guesmi and Nguyen (2011) analyse Southeastern Europe, Latin America, Asia and Middle East.

(21)

21

Rossi and Ossola, 2017). Forbes and Rigobon (2002) argue that conditional heteroskedasticity can be adjusted. They propose an augmentation, however, it is sensitive to small samples and periods when higher endogeneity is expected. To analyse whether cultural distance affects financial integration, Lucey and Zhang (2010)⁶ use both conditional and unconditional correlations.

Unconditional correlations show higher interdependence than conditional correlations, nevertheless, the results, still, are quite similar.For periods 1961-1994 (Solnik, Boucrelle and Le Fur, 1996) and 2000-2015 (Nardo et al., 2017)⁷ the results are similar – during crisis the correlation is higher. An alternative to correlation methodology is a wavelet methodology. Its peculiarity is an analysis of time series data decomposing it in frequency and time. Graham, Kiviaho, Nikkinen and Omran (2013) study interdependency of MENA countries and the United States. The results support the findings of Guesmi and Nguyen (2011) that countries are integrated regionally but not globally. Rua and Nunes (2009) find that the co-movement of major developed stock markets highly depends on the trading frequency suggesting that lower frequency increases co-movement. Co-movement methods do not evaluate market riskiness in a form of risk premium. These methods are more relevant analysing integration/connectedness trends using high frequency data. However, this methodology is not suitable to measure indirect connectedness.

Cointegration technique, proposed by Engle and Granger (1987), test whether there is a relationship between stock markets considering long-term equilibrium between the time-series which are not stationary. Palac-McMiken (1997) observe that all ASEAN markets, except Indonesia, have common long-term trend. London, Frankfurt and Paris stock exchanges are also found to be cointegrated (Kasibhatla, Stewart, Sen and Malindretos, 2006). Caporale, Gil-Alana and Orlando (2016) conclude that S&P 500 and Euro Stoxx 50 indices have unit roots, however, after the financial crisis the European Union and the United States had a different path of recovery.

Although cointegration technique is suitable assessing long-term relationships between stock markets, its drawback is restriction of analysis to one dependent and independent variable. Vector autoregression (VAR) model, instead, can assess multiple time series correlations but it does not have corrected errors (Engle and Granger, 1987), do not fit in non-linear models and not evaluate conditional heteroskedasticity (Stock and Watson, 2001). On the other hand, it captures temporal changes better than CAPM method because it is an autoregressive model. Cai et al. (2017) use this model to evaluate financial contagion in “wake-up” hypothesis where investors reassess market risk based on its fundamentals. Authors confirm this hypothesis showing that contagion can be transferred to other markets without having any financial linkages.

6 For their analysis Lucey and Zhang (2010) use daily stock market indices of 23 emerging countries.

7 Nardo et al. (2017) analyse 22 European countries.

(22)

22

Non-linearity problem is solved using generalised autoregressive conditional heteroskedasticity (GARCH) models. However, Lamoureux and Lastrapes (1990) test GARCH model analysing stock returns and conclude that GARCH model does not capture infrequent and highly irregular events causing persistence in structural shifts in unconditional variance. GARCH method is popular assessing volatility spillovers across financial markets because it considers fat tails and cluster volatility (Bollerslev, 1986). General model assumes that the residuals are normally distributed while its modifications incorporate different assumptions, for example, in EGARCH model residuals are exponentially distributed (Audrino and Trojani, 2006). GARCH model and its modifications are widely used in CAPM tests (Guesmi, Nguyen, 2014; Guesmi, Teulon, 2014;

Guesmi, Moisseron and Teulon, 2014; Carrieri, Chaieb and Errunza, 2013). Al Rahahleh, Bhatti and Adeinat (2017) using DCC-GARCH model find that equity flows between the United States and Hong Kong and the United States and Australia are bilateral. Strong unilateral correlation is found between the United Kingdom and Taiwan (UK  Taiwan) and between Taiwan and the United States (Taiwan  US). Baumöhl et al. (2018) in their analysis use GARCH and 9 its derivations. They come up with a conclusion that the highest connectedness in equity markets was in 2008 and now it is decreasing. In addition, the highest volatility spillovers come from the most liquid markets which are also the most vulnerable to volatility spillovers from other markets.

Although GARCH methodology captures volatility spillovers adjusted for heteroskedasticity, it does not evaluate multidimensional links of equity investment network.

Network methodology also has disadvantages. It is a non-parametric statistical method, therefore, analysis is based on descriptive statistics, for example, mean and standard deviation. In order to conduct statistical analysis, this method should be combined with other methods, for instance, Ordinary Least Squares (OLS) which obtain simple regressions. However, it helps to have a picture of a whole financial system and its components: central countries and peripheries, clusters and neighbours. This method is especially relevant analysing contagion spreads across the financial network and helps to detect the weak links: Sh. Zhang et al. (2016) and Chinazzi et al. (2013) implementing network methodology find that the average volume of investment significantly decreased in 2008. In addition, Chuluun (2017) comes to the results that countries which are highly integrated are also highly exposed to volatility spillovers, however, Chinazzi et al. (2013) argue that financial crisis firstly arises but also dissipates in central countries, therefore, countries in periphery are more vulnerable to negative financial shocks.

Methodologies, which assess connectedness of equity markets during growth periods, can also consider crisis impact on the level of connectedness. The reason of crisis, in other words, contagion, can be measured assessing its potential channels.

(23)

23

1.3.2 Financial contagion channels

When foreign investors diversify their portfolios in foreign equity markets, consequently, those equity markets become interconnected. However, higher financial connectedness leads to higher financial stability only until a certain level (Acemoglu et al., 2015). In addition, the risk of financial contagion depends not only on the level of financial connectedness but also on the type of links among financial markets. Therefore, there are many channels that can affect investment in different markets.

First group of researchers argues that bank deposits are spread globally and a massive withdrawal of deposits creates liquidity shocks which are transferred to other financial systems.

Allen and Gale (2000) limit their analysis to spread of contagion through banking sector assuming that investors have complete market information and there is no relation with currency markets.

They model prevalence of contagion based on the completeness of the market: incomplete international banking market with low degree of connectedness, incomplete international banking market but with high degree of connectedness and a complete market where all markets are connected. Modelling results show that complete and incomplete markets with low degree of connectedness are not contagious, while incomplete market with high degree of connectedness is susceptible to propagation of liquidity shocks. Empirical analysis done by Bekaert et al. (2014) using a factor model suggest that banking sector had no important role transferring global financial crisis of 2007-2009. In addition, different sectors were not affected homogeneously. Mollah, Quoreshi and Zafirov (2016) using adjusted conditional correlations, proposed by Forbes and Rigobon (2002), find that banking sector was the most important channel of increased correlation between equity markets. This is justified by high interconnectedness between financial sectors (Belke and Dubova, 2018). Implications of Dungey andGajurel (2015) are similar: banking sector is prone to volatility spillovers, in general, but it is found to be exposed to both systematic and idiosyncratic risks in crisis period. According an author, idiosyncratic risk can be an expression of a herd behaviour.

Herding can activate different contagion channels, therefore, contagion comes into action when risk-averse investors become even more risk averse shifting their preferences towards safer assets (Bekaert et al., 2014; Guidolin and Pedio, 2017), more liquid assets (Guidolin and Pedio, 2017) or increased risk premium (Guidolin and Pedio, 2017; Schumacher and Żochowski, 2017).

Bekaert et al. (2014) using VIX and TED spreads find that their variation increases in crisis period, however, they explain it as a measure of an econometric problem, hence, it does not reflect herding.

Other authors find mixed results (Guidolin and Pedio, 2017; Lee, 2017; Longstaff, 2010).

Analysing an Asian Crisis, Baig and Goldfajan (1999) find substantial results: when market news and economic fundamentals are controlled, shocks in other markets are transferred

Assessment of international equity investment connectedness : portfolio diversification with respect to institutional and non-institutional investors